Process for tracking and monitoring a manufacturing unit and/or a near-infrared spectrometer by means of at least one criterion of quality of sets of spectra

• US 6,012,019 A
• Filed: 10/23/1997
• Issued: 01/04/2000
• Est. Priority Date: 10/23/1996
• Status: Expired due to Fees

• Current Assignee: ELF Antar France
• Original Assignee: ELF Antar France
• Inventors: Saby, Claude-Alain
• Primary Examiner(s): Hoff, Marc S.
• Assistant Examiner(s): Bui, Bryan
• Application Number:
  US08/956,436
• Time in Patent Office: 803 Days
• Field of Search: 702/32, 702/30, 702/28, 702/27, 702/85, 250/339.07, 250/339.08, 250/339.12, 250/495.1, 250/504 R
• US Class Current: 702/32
• CPC Class Codes: G01N 21/359 using near infrared light
• CPC Class Codes: G01N 33/2829 mixtures of fuels, e.g. determining the RON-number

Abstract

A process for tracking and monitoring the operation of a unit for manufacturing a product and/or a near-infrared spectrometer fed with the product. The process includes periodically recording spectra arising from the near-infrared spectrometer, mathematically transforming the spectra, computing at least one quality criterion, intercomparing two sets of spectra after transformation, and tracking the evolution of this criterion over time. The process finds its application in the chemical, petroleum, pharmaceutical, cosmetological, and agro-food industries.

Claims

1. Process for tracking and monitoring the operation of a unit for manufacturing a product and/or a near-infrared spectrometer fed with said product, said spectrometer delivering spectra consisting of series of values of absorbance for various values of wavelengths, consisting in executing the following steps:

periodically recording, in the form of numerical data, spectra arising from the near-infrared spectrometer,

mathematically transforming the numerical data of each recorded spectrum so as to obtain transformed spectra,

characterized in that it consists in executing the following steps also;

constructing a string of work spectra from the values previously obtained, by choosing wavelengths in each transformed spectrum, by a method of selection,

selecting from the string of work spectra, a first set of work spectra containing from 20 to 50 consecutive spectra,

selecting from the string of work spectra, a second set of work spectra, of the same dimension as the first set, said spectra being consecutive and shifted in time with respect to the spectra of the first set,

computing at least one quality criterion so as to intercompare the first and second sets of spectra,

tracking the evolution of said quality criterion over time.

2. Process according to claim 1 characterized in that the method of selection of the wavelengths of the work spectra is chosen from among the method of stepwise selection, the method of selection by elimination, a method applying a regression algorithm and a method implementing genetic algorithms.

3. Process according to claim 2, characterized in that the first set of work spectra comprises spectra obtained from reference products with known characteristics.

4. Process according to claim 2, characterized in that the first set of work spectra containing a first spectrum, the second set of work spectra consists of the same spectra as the first set, except for the first which is replaced with a spectrum not belonging to the first set.

5. Process according to claim 2, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a criterion of representativity of said data, which implements a metric characterizing a distance between said sets.

6. Process according to claim 2, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a first criterion of homogeneity of said data comparing on the one hand the shapes and on the other hand the orientations of said sets.

7. Process according to claim 2, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a second criterion of homogeneity of said data intercomparing the densities of said numerical data.

8. Process according to claim 1 characterized in that, the first set of work spectra comprises spectra obtained from reference products with known characteristics.

9. Process according to claim 8, characterized in that the first set of work spectra containing a first spectrum, the second set of work spectra consists of the same spectra as the first set, except for the first which is replaced with a spectrum not belonging to the first set.

10. Process according to claim 8, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a criterion of representativity of said data, which implements a metric characterizing a distance between said sets.

11. Process according to claim 8, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a first criterion of homogeneity of said data comparing on the one hand the shapes and on the other hand the orientations of said sets.

12. Process according to one of claim 1, characterized in that, the first set of work spectra containing a first spectrum, the second set of work spectra consists of the same spectra as the first set, except for the first which is replaced with a spectrum not belonging to the first set.

13. Process according to claim 12, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a criterion of representativity of said data, which implements a metric characterizing a distance between said sets.

14. Process according to claim 12, characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a first criterion of homogeneity of said data comparing on the one hand the shapes and on the other hand the orientations of said sets.

15. Process according to claim 1 characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a criterion of representativity of said data, which implements a metric characterizing a distance between said sets.

16. Process according to claim 15 characterized in that the quality criterion is a global criterion computed via a formula involving at least two criteria chosen from among the criterion of representativity, the first criterion of homogeneity and the second criterion of homogeneity.

17. Process according to claim 1 characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a first criterion of homogeneity of said data comparing on the one hand the shapes and on the other hand the orientations of said sets.

18. Process according to claim 1 characterized in that the spectra of the first and second sets of work spectra being made up of numerical data, the quality criterion is a second criterion of homogeneity of said data intercomparing the densities of said numerical data.

19. Process according to claim 1 characterized in that in order to track the evolution over time of the quality criterion, a monovariate control chart is used.

20. Process according to claim 1 characterized in that in order to track the evolution over time of the quality criterion, a causes/effects diagram is used.